fix quoting in AsyncWgetFileGetter again

[videosite.git] / videosite / VeohGrabber.pm

diff --git a/videosite/VeohGrabber.pm b/videosite/VeohGrabber.pm
index 4cac1ea..c33f1d8 100644 (file)
--- a/videosite/VeohGrabber.pm
+++ b/videosite/VeohGrabber.pm
@@ -8,22 +8,21 @@ package videosite::VeohGrabber;
 use videosite::GrabberBase;
 @ISA = qw(videosite::GrabberBase);
 
-use XML::Simple;
+use HTML::TokeParser;
 use Data::Dumper;
-
 use strict;
 
 sub new {
     my $class = shift;
-    my $self = $class->SUPER::new();
-
-    $self->{'NAME'} = 'veoh';
-    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*veoh.com/browse/videos/category/[^/]+/watch/([^/]+))'];
-
-    bless($self, $class);
-    $self->_prepare_parameters();
-
-    return $self;
+    my $self = $class->SUPER::new(
+        NAME => 'veoh',
+        _SELFTESTURL => 'http://www.veoh.com/watch/v18348952fyn2twbe',
+        _SELFTESTTITLE => '518_2 kureyon shinchan',
+        PATTERNS => ['(http://(?:[-a-zA-Z0-9_.]+\.)*veoh.com/+watch/(\w+)\??)'],
+        @_,
+    );
+
+    return bless($self, $class);
 }
 
 sub _parse {
@@ -32,9 +31,11 @@ sub _parse {
     my $pattern = shift;
     my $content;
     my $metadata = {};
-    my $p = XML::Simple->new();
     my $ua = $self->ua();
+    my $p;
     my $t;
+    my @text;
+    my @accum;
     my $dlurl;
     my $r;
 
@@ -48,41 +49,40 @@ sub _parse {
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
-    # Get the XML file containing the video metadata
-    unless(defined($content = $self->simple_get(sprintf('http://www.veoh.com/rest/v2/execute.xml?apiKey=5697781E-1C60-663B-FFD8-9B49D2B56D36&method=veoh.search.search&type=video&maxResults=1&permalink=%s&contentRatingId=1&', $2), $ua))) {
-        $self->error('Could not download XML metadata');
+    unless(defined($content = $self->simple_get(sprintf('http://www.veoh.com/watch/%s', $2), $ua))) {
+        $self->error('Could not download');
         return undef;
     }
 
-    unless(defined($t = $p->XMLin($content, KeepRoot => 1))) {
-        $self->error('Could not parse XML metadata');
-        return undef;
-    }
+    $p = HTML::TokeParser->new(\$content);
 
-    if (exists($t->{'rsp'}->{'videoList'}->{'video'}->{'fullPreviewHashPath'})) {
-        $dlurl = $t->{'rsp'}->{'videoList'}->{'video'}->{'fullPreviewHashPath'};
-    } else {
-        $dlurl = $t->{'rsp'}->{'videoList'}->{'video'}->{'fullPreviewHashLowPath'}
-    }
+    while ($t = $p->get_tag('script')) {
+        if ($t->[0] eq 'script') {
+            my $e = $p->get_text();
+            my $jsp;
 
-    unless(defined($dlurl)) {
-        $self->error('No dlurl found in XML');
-        return undef;
-    }
+            if ($e =~ m|__watch.videoDetailsJSON = '([^\x27]+)|) {
+                $self->debug("Found JSON: %s", $1);
+                $jsp = videosite::JSArrayParser->new();
+                $r = $jsp->parse($1);
 
-    # We now have to fetch the dlurl to get the redirect target after it,
-    # because the dlurl itself must be called with the right referer set
+                unless(defined($r)) {
+                    $self->error("Found information hash, but could not parse");
+                    return undef;
+                }
 
-    $ua->max_redirect(0);
-    $r = $ua->get($dlurl, 'referer' => 'http://www.veoh.com');
+                $self->debug("Parsed JSON: %s", Dumper($r));
 
-    unless ($r->is_redirect) {
-        $self->error('Expected redirect, got %s', $r->code);
-        return undef;
-    }
+                unless(exists($r->{previewUrl})) {
+                    $self->error("previewUrl not found in information hash");
+                    return undef;
+                }
 
-    $metadata->{'DLURL'} = $r->header('Location');
-    $metadata->{'TITLE'} = $t->{'rsp'}->{'videoList'}->{'video'}->{'title'};
+                $metadata->{'DLURL'} = $r->{previewUrl};
+                $metadata->{'TITLE'} = $r->{title};
+            }
+        }
+    }
 
     unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
         $self->error('Could not extract download URL and title');